The Tower Volume 6 Issue 2

8/10/2019 The Tower Volume 6 Issue 2

1/34

GEORGIA INOFFI

225 North AvenueAtlanta, GA 30332-0325

Everywhere you look at Georgia Techtual curiosity is the launching pad for research. It starts at the undergraduateThe Towerprovides an invaluable tool the scientic community know about ing work done by Georgia Tech underg

Its crisply wrien, carefully researchenot only showcase the impressive worour campus, they also serve asinspiration for students here andelsewhere who are contemplat-ing their own research projects. Innovation and the entrepre-neurial spirit captured com-pellingly in The Towers articles are in Georgia Techs DNA.The Institute sponsors severalcompetitions that encouragestudents to undertake interdis-ciplinary research and collabo-ration as they aempt to comeup with creative solutions to de-sign, distribution, and cultural

obstacles. Our InVenture Prize,for example, encourages under-graduate students interest ininvention, innovation, and entre-preneurship.

The Tower has experienced tregrowth in recent years, both in nationand in the important audience that it rlls a valuable niche among research jostated in its Focus and Scope sectioncate the audience rather than simply sults to an expert audience in a par ticu

GPI


2/34

VOLUM

Ab o u t t he C o v er

The cover depicts an artistic visualizationof a digital circuit. The feature article ex-plores several signal-processing basedfilters to denoise electromyography sig-nals. The implementation of these filtersin hardware devices requires complexdigital circuits.

110| Prof. Eisens thoughts

on the future of publishing

and open-access science

UndergraduateResearchKaleidoscope

The Biannual Showcasing ofUndergraduate Research at the

Georgia Institute of Technology

Showcase. Present. Instruct.

gttower.org/urk/


3/34

102| Prof. Sulcheks cell sorting

microfluidic device

116| Microbial diversity of the gut

112| Flexible silicon circuit board


4/34

Focus and Scope

The Toweris an open-accreviewed research journdents of the Georgia Ingoals of our publication auate achievements in rinquiry, and promote Gto undergraduate reseapublishes articles that empirical work, includintal, historical, ethnograpary/cultural inquiry. Thereaders in academia andeasily understood by baals of any discipline. Althsubmissions of highly tetial inclusion, submissiocate the audience ratheto an expert audience insions of original researchevidence, arguments mumust be logical. Fiction, will not be accepted.

Manuscript Types

The Towerwelcomes mathe following categories Article: manuscript t

tion point of an undect, where the authfined research problmultiple approache

Dispatch: manuscripports recent progrethat is relatively narward his or her over

Perspective: manusscholarly thinking inpersonal viewpointssions on a topic of synthesis and/or log

Synopsis: manuscrioverview of an artiprofessional, refereethor must be a co-auing the research the

Submitting to The To

Submissions to The Towing basis throughout the

The TowerUndergraduate Research

Journal

ISSN: 2330-0787

Editorial Office:Office of Student MediaGeorgia Institute ofTechnology353 Ferst Drive, Suite 137Atlanta, GA 30332-0290

Contact Information:

General inquiries and theEditor-in-Chief:[email protected]

Faculty Advisor:[email protected]

Business Manager:[email protected]

Submissions & Review:[email protected]

Production Team:[email protected]

Webmaster:[email protected]

Website:

www.gttower.org

EDITORIAL BOARD

Editor-in-ChiefMohamad Ali Najia

Managing Editor of Submissions & ReviewRavi Konjeti

Managing Editor of ProductionAndrea Hu

Business ManagerReeves Broussard

Section EditorsAmrita BanerjeeJessica BlockChristopher HarperGautam Rangavajla

Layout EditorAlex Freeney

WebmasterRichard Li

REVIEW STAFF

Undergraduate ReviewersChristopher Andrew ArencibiaCasey AultmanKevin DaffonJaheda KhanamLaura LanierMatias LeguizamonAng LiDionne McKenzieMichael MerrittShamus MoranSimisola OludareChristopher PaceHifza Sakhi

Apoorv SaraogeeKunal ShahSeung Ho ShinSamantha Stadmiller

Hoki TseSarthak Vaish

Graduate ReviewersShereka BantonAmulya ChiralaNikhilesh NatrajChris QuintoShriradha SenguptaDavid SottoJenna Wilson

Post-doctoral Fellow ReviewerPritha Bagchi, PhD

Research Scientist ReviewerGaurangkumar Patel, PhD

Faculty ReviewersTibor Beseds, PhDWayne Book, PhDAmy DUnger, PhDMonica Halka, PhD

Melissa Kemp, PhDNarayanan Komerath, PhDDon Lim, PhDLakshmi Sankar, PhD

FACULTY ADVISORY BOARD

Rosa I. Arriaga, PhDJeffrey A Davis, PhDAmy DUnger, PhDMonica Halka, PhDMarlit HayslettPete J. Ludovice, PhDMilena Mihail, PhDLakshmi Sankar, PhDHan Zhang, PhD

ACKNOWLEDGEMENTS

Karen Adams, PhDCharlie Bennett

Jon BodnarRonald A. Bohlander, PhDRafael Bras, PhDBeth BryantRebecca Burnett, PhDMichael ChenStephen Cross, PhDKirk EnglehardtSteven Girardot, PhDMarlit HayslettVan JensenTyler J. KaplanKenneth Knoespel, PhDPete Ludovice, PhDCarly R. MacdonaldKathryn Meehan, PhDCarole Moore, PhDCatherine Murray-RustThomas Orlando, PhDSusan ParhamDonald MacPittsFred RascoeChristopher Reaves, PhD

Paulette A. Richards, PhDMalavika Shetty, PhDJohn ToonMichael H. YanLisa Yaszak, PhDHan Zhang, PhD

SUPPORTING ORGANIZATIONS

Auxiliary ServicesBoard of Student PublicationsLibrary Information CenterUndergraduate Research OpportunitiesProgramStudent Government Association

THE TOWERSTAFF

97 | The Tower|Volume VI, Issue II


5/34

Over the past 5000 years of human scientic exploration,the concept of large interdisciplinary research teamshas only recently emerged in an eort to solve complexscientic problems of the modern era. Mammoth scien-tic projects, where the technical goals have been clearlydened, necessitated the need for large, globally collabo-rating teams. The Human Genome Project, for example,employed engineers, biologists, chemists and computerscientists spanning four continents. Resting on the suc-cesses of the sequenced genome, the National Institutesof Health (NIH) in 2005 revealed theRoadmap Initiative to encourage, asthey term, new organizational modelsfor team science. However, have theideas of interdisciplinary science andlarge teams become inappropriatelyintertwined for todays research ques-tions?

There is no doubt that science isgeing more complex and specializeddue to the sheer volume of knowledgeproduced annually. Nonetheless, theNIH-envisioned, interdisciplinaryresearch teams of the future may notalways be required to make scienticprogress. I am not arguing against sci-entic collaborationrather, the issueis a maer of emphasis between naturalpartnerships among individual scien-tists of synergistic backgrounds and the de factoassemblyof interdisciplinary teams that reinforce old disciplines.By design, individuals within an interdisciplinary teamoften identify themselves as an expert in somethingother than the scientic problem in focus, and conse -quently abdicate the majority of scientic work becauseit is not within their eld. In this light, interdisciplin -ary teams tend to inhibit the cross-pollination of ideasbetween elds despite being comprised of individuals

from diverse disciplines. Scientic progress is drivenby questionsquestions that may demand new ways

of thinking from your prior training. You want to gowhere a question takes you, not where your train-

ing left you.When I reect upon scientic elds that have

emerged, I dont observe interdisciplinary teamscombining skills to solve a problemI observe

Interdisciplinary peopleImproving the diagnosis of neuromusculardiseases

The diagnosis of neuromuscular disor-ders, such as amyotrophic lateral scle-rosis (ALS) and multiple sclerosis (MS),is dependent on analyzing the electri-cal activity of muscles using electromy-ography (EMG). However, EMG signals

often contain high levels of background noise that candisrupt the accuracy of the electrical measurements andpotential clinical diagnoses. To address this problem, the

researchers designed a series of digital filters to excludethe problematic noise from the EMG signals. The digitalfilters were based on one of two mathematical functions,the Fourier Transform or the Wavelet Transform. Wavelet-based digital filters were determined to best excludenoise while maintaining the integrity of the EMG signal.The implications of this study will enable more accurateinterpretations of electrical muscle activity, and aid in thedevelopment of accurate neuromuscular diagnoses anddevices.

Article pg 120

The physics of table tennis

In 2000, the diameter of the tennis ballwas officially changed by the InternationalTable Tennis Federation (ITTF) to 40 mmfor international competitions in order tomake the sport more entertaining. Thesize of the ball directly influences the balls

visibility and speed, as well as the pace of game play. Com-putational simulations were performed in order to deter-mine the diameter of the ball that optimizes the duration ofgame play in which the average audience can comfortablyobserve the ball. Thus, the researchers recommend that theball diameter should be 41 mm in order to prevent unneces-sarily slowing down the game and detracting from the ballsangular momentum.

Article pg 148

How the cholera bacterium becomes

pathogenic

Worldwide, cholera affects 5 million peopleannually and is caused by the water-bornepathogen, Vibrio cholarae. V. cholaraeproduces sticky factors called biofilm toadhere to substrates in a human host oraquatic environments. At high cell densi-

ties in aquatic environments, V. cholarae represses biofilmproduction in coordination with the activation of genes

that allow the uptake of foreign DNA. The activation of suchgenes can increase the virulence of V. cholaraeand allow V.cholaraeto rapidly evolve in changing environments. There-fore, it was proposed that the accumulation of biofilm couldhinder the uptake of foreign DNA in aquatic settings. Signifi-cant decreases in DNA uptake were observed in V. cholaraestrains engineered to overproduce biofilm. Thus, this studyprovides evidence that biofilm production regulates the ca-pacity of V. cholarae to become pathogenic through the up-take of foreign DNA.

Article pg 130

Animating 2D images

Warping is the ability to create an animationfrom a two-dimensional image. Currently,there is no intuitive, customer friendly soft-ware for this process. Warpic, an Androidapplication, was created to allow the popu-lous to easily and creatively apply warping

to pictures using a touchscreen device. Users can use up tofour fingers to manipulate and contort an image in real-time,and the animation produced is the compiled motion seriesof individual warps. Warpic utilizes vector calculus to createindividual warps from user input. Warpic enables the simpleand efficient creation of image animations that can be uti-lized from the motion picture industry in generating anima-tions to the general consumer.

Article pg 140

IN THIS ISSUE

When I reflthat have em

interdiscipcollaborating

our understan



6/34

Early diagnosis of disease is always apriority in medical treatment. In gen-eral, the earlier a disease is detectedand treated, the greater the patientschance of recovery. Many diseasesincluding cancer and pathogenicconditions such as malaria result indisease-aected cells (e.g. cancer cellsor malaria-infected red blood cells)in the blood. The ablility to rapidly

detect these cells can lead to beerdiagnoses and treatments and an im-proved outcome for patients. Recentresearch from a Georgia Tech lab hasdeveloped an innovative techniquethat could isolate disease cells in a

uid medium, which may lead tonew tools for detecting diseasecells and studying disease.

The assistanWodrugineerinnew tec

based ondevice

1Wallace H. Coulter Department of BiomedicalEngineering, Georgia Institute of Technology.Correspondence should be addressed to GR (ga

Stiffness-dependent sep

cells

Gautam Rangavajla1

Recent research from a Georgia Tech lab study and diagnose disease.

Alabama Power Georgia Power Gulf Power Mississippi Power Southern Company Generation

Southern Company Services Southern Company Transmission Southern Nuclear

Southern Power Southern Telecom SouthernLINC Wireless

If youre looking for a career move that can make a real difference, consider Southern Company. Think about the fact that Southern Company

is on the leading edge of researching and developing innovative ways to provide customers with clean, safe, reliable, affordable energy.

Think about joining a company that values a culture of diversity and teamwork, where you have the opportunity to grow and develop

professionally. Think about Southern Company, where Energy, Innovation and Opportunity meet.

www.southerncompany.com/careers

Make a powermove.

2014 Southern Company


7/34

NEWS

study cancer or develop new treat-ment methods.

Dr. Sulchek sought to develop adevice capable of continuously pro-cessing and separating cells based ondierences in stiness (Fig. 1). Thedevice consists of an input for the cellsample, two outputs for the sti andsoft cells, and the separation channel,which has numerous diagonal ridges.As the uid containing the cell sampleows from the input through the sep-aration channel, the diagonal ridges

deform the cells as they ow across.

As a result of the deformation, thecells experience a transverse force rel-ative to their stiness. The stier cells

therefore experience a greater trans-verse force that pushes them upwardin the channel. The softer channelsmore easily pass through the ridges,

but are inuenced by uid ow pat-terns between the ridges, which leadto their transverse migration in theopposite direction. The outows ofthe device are situated alongside thetransverse edges (Fig. 1), and thus areable to selectively collect either sti orsoft cells.

Nanotechnology is at the forefrontof being pervasive in all aspects ofour daily lives. In the complimentaryelds of electronics and nanotech-nologies, the application possibilitiesfor the benet of mankind are vast.

Nanotechnology-based advances in-clude the potential to revolutionizemedicine, contribute to the protectionof the earths environment, enhancethe capabilities of homeland security,provide new approaches in energycreation and storage, as well as im-prove the size, performance, and ef-fectiveness of many other traditionalconsumer and industrial applications

throughout the globe. The Institute for Electronics andNanotechnology (IEN) is a large in-terdisciplinary center at Georgia Techcharged with performing cuing-edge nanotechnology research and

developing technologies that can im-prove human health, commerce, com-munications, security and the envi-ronment.

Smaller electronics, improved perfor-mance One group within the IEN is theMixed Signal Design Laboratory, di-rected by Dr. Madhavan Swamina-

The pervasive applications ofnanotechnology

Amrita Banerjee1

1School of Chemistry & Biochemistry, College of Sciences, Georgia Institute of Technology.Correspondence should be addressed to AB ([email protected]).

Dr. Sulchek believes the advantag-es over current enrichment methodsprovided by this device can at pres-ent greatly aid researchers as a labtechnique for isolating disease cells ofinterest prior to study. As a tool, theseparation device can hasten the rateof research and potentially assist in

the development of new treatments.He also hopes that this technology cansoon play a role in diagnosis by lower-ing detection thresholds for disease asan enriched sample of cells can pro-vide early notication that can lead to

beer treatment options for patients.

Research within the Institute for Electronics and Nanotechnologyis changing how humans live, communicate and treat disease.

than. The focus of his group is to min-iaturize electronic systems, with thegoal of developing new design meth-odologies that reduce the volumeof common products such as smart-phones, laptops, and super-comput-ers. Recently, Dr. Swaminathansgroup has been utilizing magneto-dielectric materials to reduce antennasize and miniaturize RF modules.They are also pioneering new electri-cal signaling techniques by integrat-ing circuits for improved power de-

livery and enhanced signal integrity.The result of these projects have

been enormous; it has resulted inmore than 50 students pursuing andcompleting their Masters and Doc-torate degrees and two spin o com-paniesJacket Micro Devices andE-System Design, says Dr. Swami-nathan. These recent achievementsaccompany several other cuing edgetechnologies and innovations devel-oped over the labs 20 year history atGeorgia Tech.

Nanoscale design of materials Dr. Seth Marders interdisciplin-ary research lab within the IEN seeks

to understand how chemical struc-tures at the nanoscale determine theelectrical and optical properties ofmaterials. Specically, the Mardergroup investigates the design of or-ganic and metal-containing materialsthat are highly polarized or polariz-able. The next generation of infor-

mation processing and communi-cation technologies will rely onthe nanoscale design of materi-

als with custom tailored elec-

trical prIn or

Dr. Mardous resmaterialhave suments, amied sin maining use

This emine to that can

ing of mto fabrices with

Needles In th10 millioyearly. Nsome phsubcutaHowevethe need

...the separation

device can hasten the

rate of research and

potentially assist in

the development of

new treatments.



8/34

NEWS

The origin of life has long been pre-dicted to have arisen from simple,self-replicating molecular systems.Many scientists have supported theRNA world hypothesis which statesthat RNA was the rst form of life onearth due to its versatility as bearersof genetic information as well as hav-ing the ability to catalyze chemical re-actions [1].

Before the evolution of DNA,RNA eectively played the role of

both DNA (genetics), and proteins,

which serve as catalysts for manybiochemical processes. However, theorigin of these biopolymers and theirevolutionary pathways was still un-clear. In a pre-biotic environment freeof proteins and enzymatic activity,the spontaneous emergence of RNApolymers remains improbable evenwith all the necessary componentsof RNA present in excess. However,researchers at Georgia Tech have pro-posed that small molecules namedmolecular midwives could possibly

serve as a mechanism for the evolu-tion of lifes earliest molecules.

Dr. Nicholas Hud, a professor inthe School of Chemistry & Biochem-istry at Georgia Tech, is the primarygure behind this groundbreakingresearch. The Hud lab is activelyworking to uncover how moleculeslike DNA and RNA rst appearedwith the origin of life on Earth four

billion years ago. The labs molecu-lar midwives theory suggests thatsmall molecules acted as templates

via non-covalent interactions with thechemical building blocks of RNA tocatalyze the formation of RNA. Thesemolecules would aid in the birthof proto-RNA, however they wouldnot be necessary once protein-basedmeans for replication evolved [2].

We envision molecular mid-wives facilitating the selection andpreorganization of the free bases byacting as nanometer-scale templatesupon which the nucleoside basesstack in aqueous solution said Hudin a publication published in the Jour-nal of Chemical Biodiversity [2].

These molecules would thereforehave to be small enough to interca-

late between the stacks of nitroge-nous bases found in DNA and RNA.In order to test this idea, Proavin, asmall compound known to bind be-tween the bases of DNA and RNA,was used as an intercalating agent forthe synthesis of a DNA double helix.Hud, along with a team of graduatestudents discovered that proavingreatly accelerates the rate at whichtwo oligonucleotides (short DNAmolecules) join into a double helix [3]

1School of Chemistry & Biochemistry, College of Sciences, Georgia Institute of Technology.Correspondence should be addressed to AG ([email protected]).

The origin & evolution of RNAMolecular midwives and their role in nucleic

acid polymerization

Alex George1

The transition from the world of RNA to DNA four billion yearsago required a helping-hand.

Figure 1 Representation of the molecular midwifehypothesis. A midwife molecule, pro-

flavin, (blue) acts as the template for the formation of a Watson-rick base pair. The removal

of these intercalating molecules leads to the formation of a DNA double helix.

(Fig. 1).With regard to the formation of

RNA, Dr. Hud theorizes that molecu-lar midwives such as proavin couldcreate a means in which nucleotidescould stack and base-pair with eachother and eventually lead to the for-mation of an RNA helix which wouldgreatly promote the formation ofmore RNA molecules. This wouldthen allow for RNA to evolve viaself-replication and eventually formDNA. These results therefore support

the labs hypothesis that molecularmidwives functioned as prebioticcofactors before the appearance of ri-

bozymes and other protein enzymes.The Hud lab plans on testing otherpotential molecular midwives inthe future and further elucidate theformation of the rst RNA moleculeswhich were critical in the evolution oflife to what it is today.

References:1. The RNAWorld: The Nature of Modern

RNA Suggests a Prebiotic RNAWorld,3rd edn., Eds. R. Gesteland, J.F . Atkins,Cold Spring Harbor Laboratory Press,Cold Spring Harbor, NY, 2006.

2. Hud, N.V., Jain, S.S., Li, X. and Lynn, D.G.

(2007) Addressing the Problems of BasePairing and Strand Cyclization in Tem-plate-Directed Synthesis, Chem. Biodiv.4, 768-783.

3. S. S.J ain, F. A. L.Anet, C. J.Stahle , N.V.Hud, Angew. Chem., Int. Ed. 2004, 43,2004.

It has beI and IIrisk of oed by hdeteriorknown tulation oment, costeoblaleads to rosis (Ri& Temen

The lab at Gethemselship betand the

ronmentions of tmarrow(MSCs),in diabelab groupotentia

betic bon

Nov

oste

Brandon

1Wallace HInstitute oCorrespon

New inosteopomedica



9/34

types in normal and high glucose en-vironments. The rst was the presenceof osteoblasts inuences adipocyte

behavior. However, the exact osteo-blast behaviors that inuence adipo -

cytes are still unknown. The next re-sult was that MSC clonogenicity andviability decreased in the presence ofosteoblasts and in high glucose con-ditions. Despite the poor viability ofMSCs in the presence of osteoblasts,the osteoblasts themselves had con-stant viability throughout the culture.This leads to the thought that MSCsact in a way to support osteoblasts indiering glucose levels at their owndemise.

The interactions observed be-tween MSCs and adipocytes dieredgreatly from the interactions between

MSCs aculture of MSCmaintengenicity

may buof clono

by hypeet al., 20in high sible thamay depthe funcdevelop

The no labthe stemFirst andnology wously stu

...cutting-edge technology w

to simultaneously study the c

to a disease condition as wel

interactions.

IBEW Local 613 ElectriciansDo Everything from Wiring Electrical Outlets

to Programmable Logic ControlsYourelectricalsystemisthelifebloodofyourbusiness. Ifitfails,yourphones

dontring,computerswontwork.YourBUSINESSSTOPS.

Dont take chanceswith your electrical system. Demand the best trained

Electrical Professionals available from an IBEW Local Union 613 Electrical

Contractor. IBEWLocal613memberscompletearigorous5yearclassroom

and onthejob training program. They know how to get your electrical

systemback

up

and

running

as

quickly

as

possible.

PowerUp. CallIBEWLocal613forthenameofQualifiedUnionElectrical

Contractors nearyou.

InternationalBrotherhoodofElectricalWorkersLocal613

501PulliamStreet,SW Suite250 Atlanta,Georgia30312

(404)5238107www.ibew613.org

MaxMountJr

President

GeneR.O'Kelley

BusinessManager


10/34

Michael Eisen is a develothe University of CalifornInvestigator of the HowaInstitute. He is also a co-f

access publisher Public (PLOS). Emajority oadvocatinthe unresto peer-research. largest peis the woculture shcommuni

The Need for Open AccesIn the 1990s, the Human Ged the focus of molecular ing individual genes in isnetworks of genes systemthe entire genome. As a pEisen was generating mas

behavior of every gene wiyeast and humans. We outo understand the results o

just by reading the literatusaid Eisen.

The explosion of scienwith the rise of the mode

beginnings of scientic joupers online. The most sensto approach the analysis

Professor Mic

Access Scienc

Professor Michael Eisen

In this special interview, E

future of scientific commu

An accomplished undergraduate research-er, Harrison Bartle graduated Georgia Techas one of the top students in the Departmentof Biomedical Engineering. He worked forthe past three years under the advisementof Dr. Lena Ting in the Georgia Tech Neu-romechanics Laboratory. His research fo-

cused on investigating human balance control through com-putational modelling and the design of robotic exoskeletons.Harrison was awarded the Best Undergraduate Poster atthe American Society of Biomechanics meeting in 2012 andco-authored a paper in the journal Gait & Posturein 2013. Heis also a recipient of the Presidents

Undergraduate Research Award,Outstanding Academic AchievementAward, and National Science Foun-dation Graduate Fellowship. Outsideof lab, Harrison served as Presidentof Delta Chi fraternity. Harrison will

be pursuing a PhD in biomechanics atVanderbilt University.

Vivian Lacayostudied BusinessAdm inist ra t ionwith a focus onhuman capitalmanagement andleadership within the Scheller College of

Business. While at Georgia Tech, she was involved in variousprofessional and cultural organizations, including the Soci-ety for Hispanic Engineers, Hispanic Recruitment Team, andAIESEC (International Association of Students in Economicand Commercial Sciences). Vivian performed research underthe advisement of Dr. Karie Davis-Nozemack, an AssistantProfessor of law and ethics. Her publication in The Toweren-titled, The Eects of the Patient Protection and AordableCare Act on Small and Medium Sized Businesses was thefeature paper of Volume VI, Issue I. Unsurprisingly, Vivianis also the recipient of the Outstanding Undergraduate Re-searcher Award from the Scheller College of Business.

An Honors Pro-gram student, Em-ily Robey-Philipsgraduated fromGeorgia Tech inMay with degrees

in International Aairs and Public Policy.She held several leadership positions on campus, includingserving as President of Sigma Iota Rho and Senior AssociateEditor for the Technique newspaper. Recently, Emily com-pleted an internship with Congressman John Lewis (GA-D)through the Georgia Tech DC Internship Program. In collabo-ration with Professor Peter Brecke, she passionately led anindependent research project on human rights, the results ofwhich were published in The Tower. Starting this fall, Emilywill begin classes at Harvard Law School. She cites that be-ing published in The Tower undoubtedly helped in the lawschool admissions process, and was an invaluable learningexperience. Emily aspires to write for one of Harvard LawSchools legal journals and become a human rights aorney.

Shannon Kehoe is a fourth year under-graduate majoring in International Aairsand Modern Languages, and minoring inEconomics. Following the publication ofher research on European immigrationtrends and aitudes in The Tower, Shan-non began to design quantitative studieson the eects of internet usage, religion, and several othermetrics on citizens aitudes towards immigration. Shannonpresented her research and represented Georgia Tech at the11th Annual Claremont-UC Undergraduate Research Confer-ence on the European Union. She was awarded a scholarship

from the Halle Foundation to intern

at the State of Georgias European Of-ce in Munich, Germany, working torecruit European companies to investin Georgia. Shannon hopes to obtaina government internship in Washing-ton, D.C. before she graduates fromGeorgia Tech.

Shannon Kehoe

Emily Robey-PhilipsVivian Lacayo

Harrison Bartlett

Previous

A u t h o r sA spotlight on four

previous authors in The

Towerwho are also some

of the brightest students

at Georgia Tech

109 | The Tower| Volume VI, Issue II


11/34


12/34

phone with exible screen, 2013). However, juhave adapted to the touch screen, a bendable scre

just be another element that most people can adju

may not be able to live without. Taking into consideration their benets anbacks, stretchable electronics may not be able tdiately eliminate their wafer-based counterpartsthicker, more rigid, and have dominated the indseveral decades. Nevertheless, they warrant refurther investigate their potential applications gaps that are beyond the reach of wafer-based

gies.

Conclusions In addition to their various promising app

one example is the pen-on-paper electronics (Russo etal., 2011). Such bendable electronics are softer than theirwafer-based counterparts, but oer only simple bendabil-ity. More recent eorts look for circuits that withstandextreme mechanics, including stretching, compressing,

bending, rolling, and twisting, which gives rise to a brandnew evolution of electronicsstretchable electronics (Kim& Rogers, 2008).

Stretchable electronics require the circuits to be ableto absorb large levels of strain (>>1%) without fracture orsignicant degradation in their electronic properties (Kim& Rogers, 2008). Unlike wafer-based electronics that arerigid and planar, stretchable electronics must be soft and

curvilinear. Stretchable electronics not only deserve ex-tensive research in the development of devices capableof conformal integration with biological tissues, which is

beyond the limit of their wafer-based counterparts; butthey also inspire a new way of thinking in the discipline ofmaterials science through the exploitation of new materialstructures.

Novel materials and new structuresStretchable electronics can be realized in two approaches.One relies on the use of novel materials in conventionallayouts, while the other relies on new structural layouts ofconventional materials (Rogers, Someya, & Huang, 2010).In the context of novel materials, the polymer/inorganicnanocomposite that displays a blend of properties of bothpolymer matrix and nano ller, is a novel class of hybridmaterial that has been widely researched in applications

of stretchable electronics. This approach requires use ofmaterials that are themselves elastic (such as polydimeth-ylsiloxane and polyurethane) to enable stretchability, aswell as inorganic nano ller (such a s silver nanowires andsingle-walled carbon nanotubes) to allow electrical con-ductivity. The elastic polymer and conductive dopants arecombined to form elastic conductors that are then used aselectrical interconnects between the devices (Rogers et al.,2010). However, all known stretchable conductors haverelatively high resistivities and properties that are highlytemperature and strain sensitive, limiting their applicabil-ity (Kim & Rogers, 2008).

A distinct approach to achieve stretchability is notthrough novel materials, but through new structural con-gurations (Kim & Rogers, 2008). The underlying idea pro-posed by Rogers et al.is very straightforward: any mate-rial that is suciently thin is exible, by virtue of bendingstrains that decrease linearly with thickness. For example,a silicon wafer (750-800 microns thick) is rigid and pla-nar, but nanoscale wires, ribbons, or membranes of silicon(100 nm - 1 micron thick) are exible (Rogers et al., 2010).Therefore, stretchability can be realized in inherently rigidinorganic materials. Compared with the novel materialsapproach discussed above, an appealing advantage of thisstrategy is that it decreases the reliance on elastic poly-

mers to accomplish stretchability, giving rise to systemswith electrical performance and reliability comparable tothose of wafer-based electronics (Kim & Rogers, 2008).

Potential applicationsThe most prominent applications of stretchable electronicslie in biomedical science and engineering, since softnessand curvedness oer matching of both shapes and me -chanical properties to biological tissues and organs (Rog-ers et al., 2010). Examples include ultrathin health-moni-toring tapes that seamlessly conform to the skin, as wellas imaging devices that u tilize hemispherical detector lay-outs to resemble an eyeball (Rogers & Huang, 2009). Theapplications of stretchable electronics have the potential torevolutionize the eld of medicine. These bio-inspired de-vices, if implemented with biodegradable materials, mightproduce far less electronic waste than would traditional

wafer-based electronics. In addition to the eld of biomedicine, a non-negligi-ble pr ot-driven market lies in consumer electronics, assuggested by Samsungs release of a phone with a bend-able screen. The commercial reality of phones with bend-able screens may bring in an unprecedented revolution ineveryday electronics leading to enormous prots for thecompanies involved. Despite the aractiveness and nov-elty of bendable screen as shown in Fig. 1, there are thosewho doubt the necessity of this feature by pointing out theconcern that it could be annoying to use a touch screenif it bends away from our nger (Samsung unveils new

PERSPECTIVE

Figure 1 Envisioned future of stretchable electronic

ploits hemispherical layout, C) electronic sensor that can



13/34

Abigail Shockey1& Frank S

The pouch m

impact on hu

1School of Biology, College o f SciCorrespondence should be addre

The human body houses bsymbionts, the majority ocial in nature. Arguably ththese symbionts are housintestine. Generally, it is udividual will part with thhabitants throughout the span; however, in certain tory bowel disease (IBD)an individual requiring ccrease. In these scenarios, aknown as total proctocoleileal pouch-anal anastomused to excise the diseased

question then arises: Howcommunity that remains cto this drastic loss of divetive highlights recent insigand addresses the next stepcurrent knowledge basis rof TPC with IPAA on the gnity.

References1. Kim, D. H., Ghaari, R., Lu, N. S., & Rogers, J. A. (2012). Flexi -

ble and Stretchable Electronics for Biointegrated Devices. AnnualReview of Biomedical Engineering, Vol 14, 14, 113-128. doi: DOI10.1146/annurev-bioeng-071811-150018

2. Kim, D. H., & Rogers, J. A. (2008). Stretchable Electronics: MaterialsStrategies and Devices. Advanced Materials, 20(24), 4887-4892. doi:DOI 10.1002/adma.200801788

3. Laplante, Philip A. (2005). Comprehensive Dictionary of ElectricalEngineering (2nd ed.). Boca Raton: CRC Press.

4. Osborne, I., Lavine, M., & Coon, R. (2010). Looking Beyond Sili-con INTRODUCTION. Science, 327(5973), 1595-1595. doi: DOI10.1126/science.327.5973.1595

5. Rogers, J. A., & Huang, Y. G. (2009). A curvy, stretchy future forelectronics. Proceedings of the National Academy of Sciencesof the United States of America, 106(27), 10875-10876. doi: DOI10.1073/pnas.0905723106

6. Rogers, J. A., Someya, T., & Huang, Y. G. (2010). Materials and Me-chanics for Stretchable Electronics. Science, 327(5973), 1603-1607.doi: DOI 10.1126/science.1182383

7. Russo, A., Ahn, B. Y., Adams, J. J., Duoss, E. B., Bernhard, J. T.,& Lewis, J. A. (2011). Pen-on-Paper Flexible Electronics. AdvancedMaterials, 23(30), 3426-+. doi: DOI 10.1002/adma.201101328

8. Samsung unveils new phone with exible screen. (2013). fromhp://www.oregonlive.com/business/index.ssf/2013/01/samsung_unveils_new_phone_with.html

PERSPECTIVE



14/34

the ileal pouchs diseased state, pouchitis. Poucchronic inammatory condition of the ileal pomay develop at any point following its constructlar in pathology to UC and CD, pouchitis may cau

bloody stools and frequent bowel movements. Tof pouchitis is still poorly understood, and an etreatment beyond the prescription of antibiotics

be discovered (Shen, 2012).While the need for further research into the p

esis of pouchitis is apparent, so too is the need ies of the eects of TPC on the functional compothe pouch microbiota. The current prevalence rais 396/10,000 (Lakatos, 2006). Twenty-four perce

(Langholz, 1994) and 70 percent of CD patients (Kim, 2012) will need surgery to control their disthat surgery is likely to consist of TPC with IPAAto understand the more nuanced eects of TPC w

both short and long-term, on the human gut micthere must be a shift in focus towards understanpouch microbiomes state of normalcy.

Complimentary to the existing phylogeneticresearch regarding the pouch microbiomes funcpacity would greatly benet the cur rent knowledThis may be accomplished using a combination onomics and metatranscriptomics. Metagenomicsdening the composition of the entire suite of sociated with a community of organisms, while mscriptomics involves determining the expressiongenes. These analyses would provide informatithe metabolic potential of the pouch microbiome

it diers from the gut microbiome i n terms of funand activity. Even more benecial, but labor intterms of sample collection, would be a longitudiof the gut microbiomes phylogenetic and functionsity at dierent time points both before and after IPAA. This would allow for the generation of a

hensive timeline of the changes that occur in thcrobiome as it transforms into the pouch micThese are just a few initial routes of researchbe taken, b ut understanding the alteratio

pouch microbiome, and subsequently the limplications of TPC with IPAA, will help

in response to this drastic loss of diversity? This perspec-tive highlights recent insights regarding this question andaddresses the next steps in expanding the current knowl-edge basis of the eects of TPC with IPAA on human gutmicrobe communities.

Microbiomes confer physiological functions to the hostWhile most microorganisms were once thought to be pri-marily pathogenic in their interactions with higher or-ganisms, it is now well understood that mutualistic asso-ciations between microbes and higher organisms are alsoprevalent in nature. In fact, almost every species of higherorganism has a consortium of microbes associated with

them, (known as a microbiome), which benets their exis-tence. The human body hosts a number of dierent micro-biomes, each varying in their phylogenetic and functionaldiversity. Microbes found within the gastrointestinal (GI)tract, known collectively as the human gut microbiome,have tenfold the number of cells and one hundredfold thegenetic potential (the number of protein-coding genes) oftheir host, with the majority of these cells concentrated inthe large intestine (Gill et al, 2006).

Recent advances in culture-independent researchmethods have shown that the gut microbiome is largelycomposed of anaerobic prokaryotes in only a few keyphyla. These phyla include Firmicutes and Bacteroidetes,as well as Fusobacteia, Actinobacteria and Proteobacteriain lower abundances (The Human Microbiome Project,2012). Although interpersonal dierences in the phylo-genetic composition of the gut microbiota are larger than

intrapersonal variation, a core set of genes and functionsare shared between microbiomes, emphasizing the im-portance of the gut microbiomes functional relationshipwith its host. Unsurprisingly, the majority of these genesare linked to the metabolism of carb ohydrates and aminoacids (Turnbaugh et al, 2009). Beyond metabolic function,the human gut microbiome is also known to promote thedevelopment and maintenance of various tissues, serve asa defense against invading pathogens and may even be ca-pable of aecting host behavior, making it a key player inhuman development, physiology and health (Felix et al,2013).

Dysfunctions of the microbiomeTPC with IPAA is a surgical procedure in which the en-tirety of the large intestine and rectum are removed fromthe gastrointestinal tract. Following the removal of thecolon, a portion of the small intestine is used to create acolon-like reservoir known as an ileo-anal pouch, whichis then aached to the patients anus to create an anasto-mosis, or union, between the two structures (Parks andNicholls, 1978). TPC with IPAA is commonly used to treatsevere cases of IBD such as Ulcerative Colitis (UC) andCrohns disease (CD). Both conditions are marked by in-ammation of the intestinal tract accompanied by fever,fatigue, nausea, cramping and bloody stools. When TPC

with IPAA is used to treat IBD, the inammation of theintestine is so severe that the colon tissue is no longer vi-able and must be excised for the well-being of the patient.This procedure has become increasingly popular over thelast few decades, because the ileal reservoir is concealedwithin the body cavity and the anal sphincter is left intact.This increases the likelihood that the patient will have ahigher quality of life following surgery (Koerdt et al, 2013). Given the sheer number of functions the gut microbi-ome performs for its host, the eects of TPC with IPAAmay extend well beyond the physical remodeling of theGI tract. Over 1014 bacterial partners are removed duringTPC with IPAA, dealing a major blow to the microbiomesdiversity and functionality (Xu and Gordon, 2003). Re-cent studies have shown that, shortly following TPC withIPAA, the new microbiota inhabiting the ileal pouch, orpouch microbiota, predominantly resembles that of the

small intestine (Hinata et al, 2012). Over time, the compo-sition of the pouch microbiome becomes more colonic innature; however, the relative abundances of individualphyla dier from those commonly found in the large in -testine. In fact, the pouch microbiome is highly enrichedin Proteobacteria, and not Firmicutes and Bacteroidetes(McLaughlin, 2010).

Normal microbiome activityBeyond its phylogenetic makeup, lile else is knownabout the healthy state of the pouch microbiome. In re-cent years, the primary focus of cli nical research has been

PERSPECTIVE



15/34

Rehman Ali1, Gareth Guvanasen2& Stephen Dew

Denoising electromyog

wavelet decomposition

IntroductionElectromyographic (EMG) signals are composetially and temporally superimposed motor unpotentials (MUAP) from various muscle motThe contractile force of an isometrically held mbe determined by the frequency of these MUA

ever, signal noise can obscure the identicati

1Wallace H. Coulter Department of Biomedical2School of Electrical & Computer Engineering, Correspondence should be addressed to RA (ra

The measurement of electromyographic (EMGmay be used to diagnose neuromuscular disocontrol electrical devices through muscle-comterfaces, or to return motor function via neuropdevices. Unfortunately, EMG measurements wcontain noise introduced by mechanical vibratitromagnetic radiation, or the electrical stimulatisue. We compare wavelet-based ltering to traFourier-based ltering methods and demonstwavelet-based approaches beer conserve EMGin the frequency domain while removing noiseable within the time domain.


16/34

by itself, cannot simultaneously render information aboutthe localized energies in the time domain and the frequen-cies expressed by the signal. An inability to process EMGregarding both of these aspects of MUAPs makes noise re-moval, and, consequently, EMG analysis dicult. There-fore, a we need a processing algorithm that perform bothtime-domain and frequency-domain analysis. A promising approach is wavelet analysis, which pro-vides information in the form of scale and time shift, asopposed to only frequency in the case of Fourier analysis.Wavelet analysis processes a signal by decomposing itsconstituent frequencies into logarithmically spaced binsin the frequency domain. Therefore, as the resolution offrequency is increased, the time resolution is reduced, andvice-versa. Despite this trade o in resolution, waveletscan be optimized for isolating time and frequency bandsto identify MUAPs of specic energies and dilations lo -calized in time. The wavelet transform has been used toremove noise from EMG signals. Wavelets such as theMexican Hat and the Morlet, in particular, have been usedto analyze EMG signals because of their resemblance toMUAPs (Laterza & Olmo, 1997).

We will demonstrate the results of noise removal bywavelet and Buerworth (a Fourier-based lter) lter-

ing on EMG data in order to determine the tradeos ofeach method on EMG. This will be accomplished in thetime domain by correlating the derivatives of the lteredEMGs (wavelet and Buerworth) and in the frequency do-main by quantifying the deviation in the derivative of thedecibel spectra for each signal. The ltering method thatdemonstrates the greatest correlation of time-domain de-rivatives and minimizes the deviation in the frequency do-main would be shown to best conserve the original EMGsignal.

MethodsTo explain our application of wavelets, we present themathematical theory behind wavelets and their applica-tion under the discrete wavelet transform (DWT). Equa-tion (1) is the basic denition of a child wavelet. The moth-er wavelet is , and the children wavelets are time-shifted(b is time-shift) and normalized dilations (a is dilation) of. The dening property of is that its integral over allspace is 0, the integral of its square over all space is 1. Be-cause the child wavelets inherit these properties, they arethe ideal nite energy signals on which a signal may beprojected. Discretization (2) of a and b into j and k enablesthe wavelet decomposition of x(t) to be expressed in series

form.

The series coecients are calculated by oprojection in equation (3) and x(t) can be reconstusing the series in equation (4). In addition to thwavelet, the scaling function (or father wavel

devised to decompose the residual function, bwould take an innite number of terms to deries (4). The key properties of are that it i

and square integrates to 1 over all space

ARTICLE

Figure 1 Decomposition Tree Up to 3 levels for the Discrete Wavelet Transform (DWT).The topmost arrow yields the approximationlevel a

3coefficients, which correspond to the V

3subspace. The rest of the arrows yield the detail levels d

1, d

2, and d

3, which correspond to

the W1, W

2, and W

3, respectively.



17/34

h[n] come from the decompositions of a mother wavelet

and a father wavelet into father wavelets that are nar-rowed by a factor of 2. This is why outputs of the ltersare then downsampled by a factor of 2 on the nodes ofthe decomposition tree in Fig. 1. The g[n] and h[n] satisfyequations (5) and (6):

(5)

(6)

According to the denitions in equations (5) and (6), the

mother wavelet denition actually comes from the fatherwavelet. Equations (5) and (6) also imply relations (7) and(8):

(7)

(8)

Furthermore, the DWT tree in Fig. 1can be reversed by

upsampling at the nodes and passing the decomthrough the reconstruction lters whose coecsimply in reverse order of the deconstruction lwavelet-based ltering method applied in this volves scaling each decomposition level to remolets of certain dilations while maintaining otherswith enhancement, and then reconstructing thwith the modied coecients by traversing thraforementioned reconstruction tree.

Stationary wavelet transform decomposition of A wavelet decomposition using the Haar wavone approximation level and ve detail levels iDWT decomposition (Fig. 2C). The mother Haa

is shown in Fig. 2Aand the corresponding fatheis shown in Fig. 2B. The d1 level is composed of

binary-narrowed (narrowed by factor of a powertime-indexed Haar wavelet. Going up the detfrom d1 to d5, the Haar Wavelet dilates by a factat each stage, making the blocks from the Haarbecome more apparent. The graphs in d

1to d

5

to zero because Haar

(t) integrates to 0, but themation level integrates to some positive nucause

Haar(t) integrates to 1. This observa

explain why the wavelet decomposition

ARTICLE

Figure 3 Example Symlet 8 Wavelet Decomposition of Raw ECG Data upto 5 levels.The red function labeled as s is the raw EMG

signal. The blue function is the approximation l evel a5. The green functions are the detail levels d

1to d

5

Figure 4 Decomposition Tree Up to 3 levels for the S

Wavelet Transform (SWT). The approximation levels a1,

correspond to the V1, V

2, and V

3subspaces. The detail leve

and d3correspond to the W

1, W

2, and W

3subspaces, respe

The filters at each level are upsamplings of the filters at thous level: g

k[n] = g

k+1[2n], h

k[n] = h

k+1[2n].



18/34

of the original EMG and consequently its derivatives af-ter ltering. Thus, to determine which ltering method

beer conserved the original EMG signal after increasingorders of dierentiation, a Spearmans correlation was cal-culated between the points of the original EMG and eachof the ltered EMGs, and between the points of each oftheir higher order derivatives as well. Furthermore, p-val-ues for each correlation values were obtained in order to

evaluate statistical signicance. The ltering method thatdemonstrates a higher statistically signicant correlationat each derivative would be deemed to best conserve thetime-domain content of the original signal.

Frequency-domain analysisTime-domain analysis provides an incomplete evaluationof how well a particular ltering method removes noise,so it is necessary to examine the frequency domain for amore comprehensive assessment. Single-sided spectrumswere generated using the Fast Fourier Transform (FFT) forthe original EMG, Buerworth-ltered EMG, and wave-

let-ltered EMGs. The amplitudes were represented ona decibel scale to accentuate details in the frequency do-main. The relative amplitude content within the frequencydomain is best described by the shape of the decibel FFTspectrum. Thus, the derivative of the amplitude in deci-

bels with respect to frequency in her was used to quan-tify shape. Root mean square (RMS) error values were cal-culated between the derivatives of the FFT spectra for theoriginal EMG and the wavelet-ltered EMG as well as be -tween derivatives of the FFT spectra for the original EMGand the Buerworth-ltered EMG. The lower error valuewould indicate which method best conserves frequency-domain information.

ResultsThe SWT-ltering and Buerworth-ltering methodswere compared in the time and frequency domain by cor-relation analysis and RMS error respectively.

Impact of the lters in the time domainThe selected segments of the original and ltered EMGsare shown in Fig. 5. We observe that at localized regionswithin the EMG signal, the stationary wavelet transform(SWT) produces a signal that best matches the shape ofthe input signal. Although the low-pass Buerworth ltereectively removes noise from the original EMG, it pro-duces a smoothed version of the original EMG and sig-nicant shape characteristics of the original signal are lost.Although the SWT reconstructed signal is slightly osetelsewhere, it retains shape characteristics beer than the

Buerworth lter.Fig. 6illustrates that for the rst derivative and high-er, the correlation of the SWT-ltered EMG to the origi -nal EMG was consistently higher than the correlation ofthe Buerworth-ltered EMG to the original EMG. Thus,SWT-ltered EMG signal beer conserves the originalEMG when taking progressively higher order derivativesthan the Buerworth-ltered EMG. Similar results wereobtained with regards to the qualitative shape character-istics and the correlations over the derivatives when theseltering methods were applied over other intervals ofEMG.

Impact of the lters in the frequency domainThe FFT spectra in decibels and the derivative ospectra with respect to frequency are shown in

Although both the Buerworth lter and thecomposition aenuate frequencies outside band of the EMG, the SWT decompositi

conserves the content of the entire freque

ARTICLE

Figure 5 Stationary Wavelet Decompostion Filtering vs But-

terworth Filter over Raw EMG Data.The green is the raw EMG sig-

nal. The red is the result of the non-causal (centered) butterworth

filter. The blue is the result of stationary wavelet transform decom-position. The selected region of EMG was analyzed and shown.

Figure 6 Stationary Wavelet Decompostion Filtering

terworth Filter over Raw EMG Data.The blue shows the

tion between the stationary wavelet transform decompothe original EMG. The red is shows correlation between th

causal (centered) butterworth filter and the original EMG

All correlations shown in (B) were calculated for the plots

Furthermore all significant correlations (> 0.01) were bols

< 0.05.



19/34

and is used to make progressively beer approximationof x(t). The element that adds more detail to V

jis the child

wavelet narrowed by factor of two: j-1,k

(t)=2j,k

(2t). Thus,V

j-1 is composed of V

jand the incremental level of detail

Wj, which is the subspace of wavelets of a certain dilation.

(9)

Convergence thus depends on diminishing levels ofdetail coming from W

j as j increases. The recursive set

denition from (9) means the following after iterative dila-

tions in discrete time:

(10)

(11)

Because of sampling from continuous time to discretetime, the wavelet cannot be innitely narrow; therefore,W

1denotes the level of the most narrowed wavelet in dis-

crete time. The set notation (10) describes the Wavelet de-composition performed up to n levels. In Fig. 2and Fig. 3,the approximation level a

5correspond to V

n, and the detail

levels d1to d5correspond to W1to Wn The nth level re-sidual or the approximation level Vncan be decomposed

into the scaling functions n,k

(t). If n=, Vn would be the

DC component of a periodic signal.These decomposition properties indicates that if wave-

let decomposition were not carried out to the sucientnumber of levels, V

nwould present a signicant amount

of baseline wander. Because the windows of analysis were1024 sample long, the largest n could have been was 10 (210= 1024); thus V

10still represented a signicant amount of

baseline wander. The only way to ameliorate this baselinewander is to expand the analysis window exponentially;however, this window expansion would lead to signicantincreases computation time. Thus, our choice of n = 10 was

based on striking the balance between the convergence ofL2(R) and minimizing computation time.

Characterizing shape by derivativesSoftware packages that employ wavelet decompositionon EMG have been shown to identify and extract indi-vidual MUAPs based on their shape characteristics (Ze-nnaro, Wellig, Koch, Moschy, & Laubli, 2003). Becausethe shape of the original EMG is best dened by its deriva-tives, a Spearmans correlation was conducted betweenthe points of the original EMG and the ltered EMGs (Fig.6). It was discovered that for the rst derivative and high-er, the correlation of the SWT-ltered EMG to the originalEMG was consistently higher than the correlation of theBuerworth-ltered EMG to the original EMG. This pres-

ervation of time-domain derivatives may be of granalyzing muscle mechanics through EMG becauliterature we nd that neural activation of muscdened by the derivatives of the EMG signal ratthe EMG signal itself (Buchanan, Lloyd, Manal, 2004).

Our method of determining the ecacy of deviates from the convention, which involves mithe square of the error and having the wavelet dsition provide beer convergence to the origin(Daubechies, 1990). This convention, however, provides information regarding the conservatioderivatives of the EMG. Furthermore, it assumes

vergence to the original signal determines ecamethod. It is because wavelets conserve the keywithin a signal that methods such as the waveletal ANOVA have been developed to identify stsignicant dierences between EMGs (McKay, Wdakovic, & Ting, 2013).

Characterizing shape in frequency domainThe frequency domain was examined because main analysis alone remains an incomplete evalhow well a particular ltering method removes such, frequency-domain analysis was used to supthe aforementioned time-domain analysis. It waered that the SWT beer conserved the frequencof the original EMG while aenuating the frequeside the baseband. The weights given to the leveto d

10compress the signal at certain scales (or d

rather than frequencies. As a result, wavelets eto remove noise at certain frequencies without rEMG information at those frequencies. Furthermscale is related to logarithmically spaced bins inquency domain, wavelets can also be exploited tate a signal at certain frequencies just a s in this w

Future WorkThe work presented herein does not provid

plete analysis of Fourier-based and waveltering techniques for analyzing EMGFurther research should aempt to dem

ARTICLE

Figure 7 Stationary Wavelet Decompostion Filtering vs But-terworth Filter over Raw EMG Data.The green is the raw EMG

signal. The red is the result of the n on-causal (centered) butter-

worth filter. The blue is the result of stationary wavelet transform

decomposition. In this figure the whole E MG signal was analyzedfor its spectrum (determined by Fast Fourier Transform) and the

amplitude is plotted on a decibel scale (A). A derivative of the deci-

bel spectrum with respect to frequency (B) was take to analyze the

shape of (A).



20/34

Sarah Wilson1, Eryn Bernardy1& Brian Hammer,

Biofilm production regu

Vibrio cholera

IntroductionMany bacteria use a form of chemical commutermed quorum sensing (QS), to regulate gene ein response to population density. This is accobecause the bacteria make and secrete chemic

called autoinducers (AIs). When the populaties a high enough density, or a quorum, the

1School of Biology, College of Sciences, GeorgiaCorrespondence should be addressed to AS (ash

The waterborne bacterial pathogen Vibrio cholizes a cell-cell communication system calledsensing to coordinate group behavior in both host and aquatic environments. Virulence genecholera toxin, biolm genes for sticky secretement factors, and competence genes for DNA uall regulated through this population density-dsystem. In a human host, both virulence andgenes are repressed late in infection presumabmote transmission upon exhausting the hostses. However, in the natural environment, regumore complex. Namely, at high cell densitiession of biolm production is coordinated with a

ARTICLE

6. Liu, H. H., Chen, X. H., & Chen, Y. G. (2005). Wavelet transformand real-time learning method for myoelectric signal in motiondiscrimination. 7th International Symposium on MeasurementTechnology and Intelligent Instruments, 13, 250-253. doi: Doi10.1088/1742-6596/13/1/058

7. McClellan, J. H., Schafer, R. W., & Yoder, M. A. (2003). Signal pro-cessing rst. Upper Saddle Ri ver, NJ: Pearson/Prentice Hall.

8. McKay, J. L., Welch, T. D. J., Vidakovic, B., & Ting, L. H. (2013). Sta-tistically signicant contrasts between EMG waveforms revealedusing wavelet-based functional ANOVA. Journal of Neurophysiol-ogy, 109(2), 591-602. doi: DOI 10.1152/jn.00447.2012

9. Misiti, M. (2007). Wavelets and their applications. London ; New-port Beach, CA: ISTE.

10. Santoso, S., Powers, E. J., & Grady, W. M. (1997). Power qualitydisturbance data compression using wavelet transform methods.IEEE Transactions on Power Delivery, 12(3), 1250-1257. doi: Doi10.1109/61.637001

11. Zennaro, D., Wellig, P., Koch, V. M., Moschy, G. S., & Laubli, T.(2003). A software package for the decomposition of long-termmultichannel EMG signals using wavelet coecients. IEEE Trans-actions on Biomedical Engineering, 50(1), 58-69. doi: Doi 10.1109/Tbme.2002.807321



21/34

ARTICLE

cholerae, responsible for the fatal cholera diarrheal disease,QS controls numerous genes that are critical for interac-tion with a human host and survival in a marine environ-ment. However, quorum sensing alone is not sucientfor V. choleraeto regulate certain genes in aquatic seings.Specically, to express genes to become naturally compe-tent and take up extracellular DNA, one mechanism ofhorizontal gene transfer, V. choleraerequires both QS AIsand the presence of chitin, which is a polymer of the sugarN-acetylglucosamine and the material that comprises theshells of many aquatic organisms (Sun et al. 2013).

Horizontal gene transfer (HGT) is the process bywhich bacteria are able to incorporate foreign genetic ma-

terial onto their chromosome and acquire new traits to bet-ter adapt to their environment. The DNA acquired couldinclude genes for antibiotic resistance, virulence factors,or novel metabolic capabilities that are particularly usefulfor surviving harsh environments (Sun et al.2013). QS notonly controls natural competence in V. cholerae, but alsomany other phenotypes including virulence and biolmformation (Zhu et al. 2002; Hammer and Bassler, 2003).While numerous genes are activated and repressed by QSin response to cell density (Ng and Bassler, 2009), it re-mains poorly understood how this timing contributes tomany QS-controlled behaviors. Previous studies (Ng and Bassler 2009) have shownthat at low cell density (LCD), when AI levels are low, theQS pathway results in the expression of virulence genes,such as the ctxgene for the cholera toxin, as well as vpsgenes which synthesize Vibrio polysaccharides crucial inthe production of biolm (Zhu et al. 2002; Hammer andBassler 2003). Conversely, at high cell density (HCD) in thepresence of chitin, when autoinducer levels are high andvirulence and biolm genes are repressed, genes codingfor factors comprising an apparatus for taking up extra-cellular DNA are activated. These genes are regulated byHapR, which acts as both an activator and a repressor ofgene expression (Bardill et al. 2011). The proposed expla-nation for why V. choleraeuses quorum sensing to turn ogenes when in a group, in contrast to other pathogens likePseudomonas aeruginosa that turn on genes when at highdensities, is that V. cholerae thrives by entering a human

host, replicating, and exiting into the aquatic environmentto nd a new host (Hammer and Bassler, 2003; Srivastavaet al. 2011). Therefore, instead of waiting to aack the hostuntil HCD has been reached as in other species, the down-regulation of biolms and virulence at HCD allow for atimely exit by V. choleraeafter maximally utilizing the re-sources within the human host. However, distinct from a human host, in aquatic set-tings where V. choleraeis commonly found, this model forthe timing of factors is no longer relevant. V. choleraecom-munities are typically found on chitinous surfaces such ascrab shells (Meibom et al.2005). HGT can be accomplished

by transduction via viral infection, by conjugation where

direct contact between two bacterial cells allows for directpassage of genetic material via a sex pilin, and by naturaltransformation where bacteria acquire naked DNA fromtheir environment. Until recently, natural competence byV. choleraehad not been demonstrated, but by studying thenatural environment in which V. cholerae commonly re-sides, it was discovered that chitin itself serves as a signalto induce the natural competence genes of V. choleraere-quired to take up DNA (Meibom et al. 2005). The presenceof chitin induces the expression of the TfoX regulator andas discussed, QS promotes production of HapR. Both ofthese regulators have been shown to be required for acti-vation of numerous DNA uptake genes including comEA,which encodes a component of a DNA uptake apparatus(Antonova and Hammer, 2011). The impact of biolms onthe process of HGT i s not yet understood, though produc-tion of biolm appears deeply intertwined with both path-ways leading to competence. Additionally, bacteria often produce many diguanyl-ate cyclase and phosphodiesterase enzymes which workto produce and degrade c-di-GMP, a secondary signalmolecule which promotes the expression of aachmentfactors and biolm formation, respectively (Massie et al.2012; Zhao et al.2013). HapR accumulation at HCD inhib-its c-di-GMP production and the loss of this small mole-cule prevents the transcriptional activators VpsT and VpsRfrom inducing expression of the vps genes (Waters et al,2008; Srivastava et al, 2011). Thus, HapR accumultion atHCD leads to repression of biolm formation concomitant

with activation of competence genes for DNA upThe levels of c-di-GMP in the cell can be ex

tally controlled by genetic modication to V. chsulting in alterations in the amount of biolm p(Srivastava and Waters, 2012). In this study suchulations are exploited to test the impact of bi

cumulation on natural competence for DNAInnappropriate formation of biolm was testtermine whether it can hinder uptake of ext

DNA and whether the production of biolfully regulated to maximize the potential uptake in the natural environment whil

Figure 1 Model of the QS-dependent regulation of c

densities.



22/34

ARTICLE

can alter the capacity of V. choleraeto engage in transfor-mation, and whether upregulation of competence geneswith the downregulation of biolm production by the sig-naling network (Fig. 1) maximizes the potential for DNAuptake. Therefore, unlike wildtype V. choleraestrains thatproduce biolms at LCD (Fig. 1A) and then stop express-ing them at HCD (Fig. 1B), V. choleraemutant strains thateither make no biolms or make biolms constitutivelyat both LCD and HCD were engineered. Specically, V.choleraemutants were constructed, biolm formation wasmonitored, and the frequency of DNA uptake by V. chol-eraewas calculated under a variety of dierent physicaland time-dependent conditions.

Genetic engineering of mutant V. cholerae strainsTo uncouple upregulation of competence genes and down-regulation of biolm genes by QS (Fig. 1), a V. choleraevps

deletion mutant was created that is unable to form biolms(vps) using standard allelic exchange methods. A con-stitutive biolm-producing strain (hapR) was also engi-neered that makes biolms at LCD a nd HCD, irrespectiveof QS function. Specically, plasmid pCMW75 describedpreviously (Waters et al, 2008) was used, which can be in-duced with IPTG to express a diguanylate cyclase previ-ously shown to activate vpstranscription and promote bio-lm formation, independent of QS. The pCMW75 plasmid(labeled as DGC vector in Fig. 2) was introduced intothe various V. choleraestrains tested. An empty vector thatdoes not carry the diguanylate cyclase gene for c-di-GMPproduction was used as a control. Biolm production wasmeasured by a standard crystal violet staining biolm as-say (Hammer and Bassler, 2003). These strains and meth-ods were used to determine how the dierent genotypesaect formation of biolms.

DNA uptake eciencyThe standard assay used to quantify the eciencuptake by V. choleraeinvolves incubating the bactures in articial seawater on a chitinous crab sto induce natural competence (Meibom et al, 20

bacteria are then incubated for an additional with exogenous DNA (eDNA) marked with theresistance to the antibiotic kanamycin (kanR). Fcholeraeare plated onto LB medium containing oantibiotic. The eciency of DNA uptake is calcthe number of colonies that grow on the selecti

(LBkan) divided by the number of colonies onsive plates (LB). This assay quanties how ebacteria take up DNA, a nd was used to d

whether deciency or prociency in bioltion alters this process.

Figure 2 Biofilm

assay mean optical

densities+/- standard

deviation of eightreplicates of wild

type (WT),vpsR, andhapRstrains con-

taining control andinducible diguanylate

cyclase (DGC) overex-

pression vectors. * =

significant differenceof transformation

frequencies from the

wild type (WT).


ARTICLE


23/34

ARTICLE

type strain C6706 (WT) produced reasonable biolms asdescribed prior (Hammer and Bassler, 2003), and the addi-tion of IPTG did not eect biolm levels (Fig. 2, bars 1-2).There was a decrease in biolms for the WT strain withthe control vector, but this is likely an eect of the added

antibiotic required to maintain the vector, and addition ofIPTG had no eect (Fig. 2, bars 3-4). The WT strain carry-ing the DGC vector (pCMW75) was similar to WT whennot induced with IPTG, but showed a signicant increasein biolm when IPTG was added (p


24/34

ARTICLE

changes for a timely exit from the infected host. However,V. choleraespends the majority of its life in marine envi-ronments and the need for this precisely regulated systemwas unclear in this environment. This study revealed asimilarly balanced relationship as the one described above

between biolm production and DNA uptake. Overproduction of biolm through the use of an in-ducible plasmid signicantly decreases the transforma-tion frequency of V. cholerae (Figure 3). However, addi-tional work needs to be done to determine whether thisresult is due to the accumulation of biolm material onthe outside of the cells impeding DNA uptake or other is-sues with overexpression of a diguanylate cyclase. For ex-ample, it is possible that accumulation of c-di-GMP insidethe cells eects DNA uptake in a manner that is unrelatedto biolm production. It is known that c-di-GMP insideV. choleraecells not only binds to VpsT and VpsR to alter

biolms, but in other bacteria also binds numerous otherfactors (Hengee, 2009). Even if the eects were due to theaccumulation of c-di-GMP, it would be interesting to seeif c-di-GMP alters DNA uptake as this has not been docu-mented to date. It was also seen that V. choleraemutants (vpsR) thatproduce no biolms have a reduced transformation fre -quency compared to WT C6706 (Fig. 4). Therefore, it seemslikely that not only does an overproduction of biolm af -fect the ability of V. choleraeto take up extracellular DNA,

but minimal biolm also impairs competence. Hence, itmay be benecial for V. choleraeto tightly regulate the bal-ance of biolm formation and DNA uptake through thequorum sensing pathway. In future works, directly testing

the expression levels of biolm genes in response to theoverexpression plasmids will ensure that the seen eectsare not due to other eects of the vector inside the cell.There should also be more work done in co-culture assayswhere extracellular DNA is provided by donor cells in theassay, to test the eects of biol m formation when V. chol-eraeis interacting with other bacteria as in nature, as op-posed to articially adding extracellular DNA. There arestill many unanswered questions on how biolm is able toaect the horizontal gene transfer of V. cholerae. This studyis a step towards fully understanding the evolutionary ad-

vantages of maintaining the quorum sensing sys tem in na-ture which, when introduced into human hosts, regulatesthe virulence of V. cholerae.

AcknowledgementsThis work has been supported by a grant from the NSFand by a PURA grant from the Georgia Institute of Tech-nology. I would also like to thank my PI, Brian Hammer,for his support and discussions and my graduate studentmentor, Eryn Bernardy, for all of her guidance.

References1. Antonova, Elena S., & Hammer, Brian K. (2011). Quorum-sensing

autoinducer molecules produced by members of a multispeciesbiolm promote horizontal gene transfer to Vibrio cholerae. FEMSMicrobiology Leers, 322(1), 68-76.

2. Bardill, J. P., X. Zhao & B. K. Hammer, (2011) The Vibrio choleraequorum sensing response is mediated by Hfq-dependent sRNA/mRNA base pairing interactions. Molecular Microbiology 80, 1381-1394.

3. Berk, Veysel, Fong, Jiunn C. N., Dempsey, Graham T., Develioglu,Omer N., Zhuang, Xiaowei, Liphardt, Jan, Yildiz, Fitnat H., & Chu,Steven. (2012). Molecular Architecture and Assembly Principles ofVibrio cholerae Biolms. Science, 337(6091), 236-239.

4. Davies, David G., Parsek, Mahew R., Pearson, James P., Iglewski,Barbara H., Costerton, J. W., & Greenberg, E. P. (1998). The In -volvement of Cell-to-Cell Signals in the Development of a BacterialBiolm. Science, 280(5361), 295-298.

5. Hammer, Brian K., & Bassler, Bonnie L. (2009). Distinct SensoryPathways in Vibrio cholerae El Tor and Classical Biotypes Mod -ulate Cyclic Dimeric GMP Levels To Control Biolm Formation.

Journal of Bacteriology, 191(1), 169-177.6. Hammer, Brian K., & Bassler, Bonnie L. (2003). Quorum sensing

controls biolm formation in Vibrio cholerae. Molecular Microbi-ology, 51(5), 1521-1521.

7. Hengge, Regine. (2009). Principles of c-di-GMP signaling in bacte-ria. Nat Rev Micro, 7(4), 263-273.

8. Massie, Jonathan P., Reynolds, Evan L., Koestler, Benjamin J.,Cong, Jian-Ping, Agostoni, Marco, & Waters, Christopher M.(2012). Quantication of high-specicity cyclic diguanylate signal-ing. Proceedings of the National Academy of Sciences.

9. Marvig, Rasmus, & Blokesch, Melanie. (2010). Natural transforma-tion of Vibrio cholerae as a to ol - Optimizing the procedure. BMCMicrobiology, 10(1), 155.

10. Meibom, Karin L., Blokesch, Melanie, Dolganov, Nadia A., Wu,Cheng-Yen, & Schoolnik, Gary K. (2005). Chitin Induces NaturalCompetence in Vibrio cholerae. Science, 310(5755), 1824-1827.

11. Ng, Wai-Leung, & Bassler, Bonnie L. (2009). BacteriaSensing Network Architectures. Annual Review of Gen197-222.

12. Srivastava, D., Harris, R. C., & Waters, C. M. (2011).of cyclic di-GMP and quorum sensing in the control oaphA in Vibrio cholerae. J Bacteriol, 193(22), 6331-6341

13. Srivastava, D., & Waters, C. M. (2012). A tangled webconnections between quorum sensing and cyclic Di-GMriol, 194(17), 4485-4493.

14. Sun, Yan, Bernardy, Eryn E., Hammer, Brian K., & Miya(2013). Competence and natural transformation in vibrilar Microbiology, 89(4), 583-595.

15. Waters, C. M., Lu, W., Rabinowi, J. D., & Bassler, BQuorum sensing controls biolm formation in Vibr



25/34

Brian Edmonds1& Jarek Rossignac, PhD2

Warpic: A four-finger sys

dimensional image warp

1School of Computer Science, College of Comp2 School of Interactive Computing, College of CoCorrespondence should be addressed to BE (br

Image warping can be dened as the manipuportions of an image through user input. Wea software system called Warpic that allows foation of two-dimensional animations from statvia user controlled image warping. The initial of the users four ngers are recorded when the ues them on a touch screen device. As the user drngers over the screen, Warpic computes a spfor each frame and applies it to the image. An application was built in order for users to creattions using four ngers as simultaneous inputsbile device. A back-end web server was implemusers to upload and share their animations. Thprovides a unique user interface that humans cnavigate. Additionally, the Warpic software alloto create most animations in real-time, without spent waiting for rendering or conguration.

Meet Josh.Upon graduation, Josh joined ourInside Territory Sales ManagementTeam. The KORE culture fosters rapiddevelopment and he capitalized upon that

and received several quick promotions.Josh is now the youngest - and one of themost successful - members of our BusinessDevelopment Team.

Your success awaits

As a leading Atlanta technologycompany, we create successstories for the graduates we hire.

Contact us today to see howKORE can help you succeed too,at www.koretelematics.com/careers.

Expand Your Opportunities

Acuity Brands, Inc. is a North American market leader

and one of the worlds leading providers of lighting

solutions for both indoor and outdoor applications. Our

innovative lighting solutions cover both conventional

fixtures and advanced solid-state technology that can

seamlessly integrate with powerful digital controls and

daylighting to provide greater energy efficiencies and ahigher quality of light to our customers.

We are seeking talented Georgia Tech graduates

interested in working in a challenging, fast-paced, and

collaborative environment with many career

opportunities for professionals that are ready to perform.

Currently, we have openings for challenging roles within:

Engineering/Product Development

Supply Chain

IT

Sales & Marketing

Leadership Development Programs in Business,Engineering, and Supply Chain

For more information, please visit our Career Center at:

http://www.acuitybrands.com/acuitybrandscorporate/

home/about-us/careers

ARTICLE


26/34

ARTICLE

manipulating real objects. Warpics user interface must beintuitive and must have dynamic and versatile abilities,meaning that the user can specify many dierent anima-tions and create the desired acceleration eects. Accelera-tion eects refer to the speed at which certain portions ofthe animation move at once. Several problems make accomplishing all of thesegoals challenging. Specifying user input to generate im-age warps in an ecient and easy manner is dicult. Thetraditional point and click interface requires time andmuch trial and error from the user when specifying manydierent control points. Additionally, keeping a warpinganimation spatially smooth is a challenge. However, muchprogress has been made on image deformation challenges.Schaefer and McPhail developed a strategy, Moving LeastSquares (MLS), for deforming two-dimensional imagesin 2006 [1]. This work succeeded in generating spatiallysmooth warps as well as calculating resulting warps in re-al-time. Also, MLS is not restricted to four control points.The warped image is dened as a weighted average ofthe current position of the control poi nts. The weights arecomputed using the distances of the optimized points toensure that the warp interpolates the control points. Maand Zhao expanded on this technique with their paperMoving Regularized Least Squares [2]. The deformationscheme involves the user choosing a set of handles anddestinations for these handles. The handles can be edgesor points.

Several dierent techniques exist to generate anima -tions. Key frame animation is used often in the animationpipeline, however the user often must specify many dif-

ferent frames to build a substantial animation. When us-ing key frames, creating expressive animations with sud-den accelerations or subtle movements may be tedious.Furthermore, this approach may require designing andediting a large number of key frame curves. In contrast,puppeteering, a form of motion capture and popular alter-native to key framing, captures detailed movements eas-ily. Puppeteering does have two drawbacks; it does notsupport editing and it limits the number of parametersthat can be controlled by the user. Relevant work exists that utilizes puppeteering to

control 2 and 3 dimensional virtual characters within acomputer. For example, Shitori and others at MicrosoftResearch allow the user to move a control piece in 3 di-mensions, hence creating puppeteering animations [3].They created a system that oers real-time feedback andprovides a much quicker and exible interface than keyframing. Although 3D sensing devices make it possible totrack users hands and ngers, we focus on touch screentablet interfaces and limit the tracking to four ngers. Theuser can maintain a real-time connection with their anima-tion and have ne-grain control of animation dynamics.The captured animation replays at the speed by which theuser moves his or her hands. This input scheme generatesa predictable, natural and simple output.

When choosing the number of ngers to accept as in-put, a few previous studies helped guide our decision touse four ngers. In 1986, Buxton and Myers published apaper stating that bi-manual interfaces in computing arenatural for humans [4]. With two hands, the user can eas-ily create symmetrical eects within his or her animations.Moscovich and Hughes studied the eects of dierent in-put mappings for multi touch inputs, noting that often awidget or component present on touchscreens is smallerthan a persons actual ngers[5]. With this in mind, we de-signed Warpic with a teleport interface, meaning Warpicallows the user to place ngers on the screen anywhereand avoids the necessity to grab specic handles on thescreen.

Methods

User workow for animation generationWarpic does not require any complicated seings nor doesit require the specication of any key frames or animationscurves. The user simply starts with a static image that can

be loaded from a le or taken with the tablets camera (Fig.1). Upon choosing an image the user can beginpre-warpingthe image. The user places four ngers on the screen andwarps the static image into a starting point for motion cap-ture. Next, the user species an animation through pup-peteering, a simplied version of motion capture in whichthe user moves four ngers on the touch screen (Fig. 1).

Finally, Warpic calculates the ensuing animationIn Warpic, the initial positions of the four ngecorded when the user places them on the toucThen, as the user drags the ngers over the screeframe, Warpic computes a space warp and ap

the image. Once the user lifts all four ngers tcapturephase begins. The motion capture lows the user to record a motion path of fou

while showing the warping animation

Figure 1 User workflow for generating animations in

(middle) to create a warped, two-dimensional animation,


ARTICLE


27/34

ARTICLE

Finger tracking algorithmWe designed a four-nger interface that has the sole re -sponsibility of intercepting the Android motion eventscreated by the users ngers on the scr een. Motion eventsare the system data that result from dierent user input.A challenge with this includes correctly indexing the datathat is received as the user lifts, drags and touches thescreen with multiple ngers simultaneously. Due to thefact that the frame rate of animation must be constant butthe user creates motion events at an unpredictable rate,our interface queues these events so that they can be han-

dled at a set point in the draw loop.

Vertex and grid formationAt every frame, Warpic denes a warp. The software cre -ates a grid beneath the image and the vertices of this gridare displaced according to the users input as interpreted

by our four-nger interface. Performance of Warpic bole-necks at the resolution of this grid. As the resolution of thegrid increases, the ability of the tablet to maintain its real-time connection with the user falters. One can begin to seenoticeable lag in the users actions and the displayed warp

when the grid is 100 vertices by 100 vertices. For testingpurposes the grid resolution stays at 50 X 50.

The four ngers of user input is mapped to two edg -es L and R, where a us ers thumb and forenger of theleft hand are points on the x,yplane L

A and L

B( Fig. 2).

Predictably, a users thumb and forenger from the righthand are (x,y) points referred to as R

A and R

B. Warpic

keeps track of the positions of these edges when the userrst puts his or her hands on the screen as L

0and R

0(Fig.

2). Warpic also keeps track of the current position of theedges at the current moment in time as L

tand R

t.Finally

the image is painted over the grid using texture mappingbased on the warp calculated from inputs Land R. During

the motion capture phase, the user moves his orngers around on the screen and a t every frametion to calculating a new warp, the values of LA

t,

RBt are added to history lists. The action his

represent a motion path dened by the user andated over during animation playback.

Calculation of warpGiven an edge between the users forenger an

consider the following points:A

0: the rst location that the thumb touched the s

At: the current location of the thumb

B0: the rst location that the forenger touched th

Bt: the current location that the forenger tou

screen.

The software follows the below steps to calcwarp at every frame, which essentially is a ment of vertices (Fig. 3).

Figure 2 Underlying grid during warping state. L0and R

0are

shown in red and represent the initial placement of the users four

fingers, while Ltand R

tare shown in blue and represent the posi-

tion of the users four fingers at the current moment in time.

Figure 3 Spiral motion formation with two fingers.The users

thumb and forefinger, shown as Aand Bat different points in time,

are used to generate a family of logarithmic spirals that converge ata spiral center, C.

Figure 4 Adjustable region of influence for image w

(r/R) which is visualized to the u ser with the green circle.

>1. With this adjustment, the new finger proxies are show

magenta edges.


ARTICLE


28/34

ARTICLE

3. Calculate point Cas the convergence of a family of spi-rals that also intersect at A

0, A

t, B

0and B

t. The software

uses the stroke motion method outlined in the Rossignacand Whiteds paper, InBetweenIt [7].

4. Next rotate a given point P, around the spiral center C,by the calculated angle agiving us point P.

P= C + satCP

Pis then linearly interpolated towards Cwith our calcu-lated spiral scale, s(Fig. 3).

Warpic follows these steps for both hands. Displacing thevertices in the grid once for each hand, essentially blend-ing the eects of both hands. A

Documents

The Tower Volume 6 Issue 2